Diffusion-weighted imaging-derived parameters as quantitative imaging biomarkers in magnetic resonance-guided radiotherapy: a systematic review.

Background and purpose: Magnetic Resonance Linear Accelerators (MR-LINACs) have transformed radiotherapy by integrating high-resolution magnetic resonance imaging (MRI) with precise radiation delivery. Diffusion-Weighted Imaging (DWI)-derived parameters are promising non-invasive quantitative imaging biomarkers (QIBs) for MR-guided radiotherapy (MRgRT), primarily through apparent diffusion coefficient (ADC) and intravoxel incoherent motion (IVIM) models; however, their clinical utility and validation on MR-LINAC systems remain underexplored. This systematic review evaluates DWI's technical development, validation, and clinical role in MRgRT using MR-LINACs. Material and methods: Following PRISMA guidelines, PubMed/MEDLINE, Web of Science, and Scopus (2014-2024) were searched for English-language studies on DWI in MRgRT with MR-LINACs (0.35 T or 1.5 T). Eligible studies included technical and clinical research with phantoms, volunteers, or patients. Data on study characteristics, DWI protocols, validation metrics, and clinical endpoints were extracted and qualitatively synthesized; heterogeneity precluded meta-analysis. Results: Thirty-four studies (11 at 0.35 T, 23 at 1.5 T) were included (29 prospective, 5 retrospective), with 28 involving patients (N = 484) across cancers. DWI, primarily via single-shot EPI (median time ∼4-5 min, b-values 0-2000 s/mm2), demonstrates robust technical feasibility (27 studies) and emerging clinical validity (8 studies, ∼212 patients). Most validation remains single-center; multi-center and cost-effectiveness data are lacking, and only one study systematically evaluated imaging-genomic correlation with IVIM. Discussion: DWI on MR-LINACs is technically available, feasible, and clinically promising for MRgRT. However, its QIB potential is limited by inconsistent protocols, hardware constraints, and preliminary clinical validation. Standardized protocols, hardware upgrades, and rigorous multi-center trials are essential to establish DWI-derived parameters as reliable QIBs for MRgRT.